JTTEE5 15:364-371

DOI: 10.1361/105996306X124365
1059-9630/$19.00 © ASM International
Peer Reviewed

Measurements of Cold
Spray Deposition Efficiency
S.V. Klinkov and V.F. Kosarev

(Submitted September 7, 2005; in revised form January 18, 2006)

An analytical model of the kinetics of coating formation during cold spray is presented. The model is used to
correct experimental data on deposition efficiency. The experimentally observed values are shown to be
affected by experimental conditions, such as the velocity of substrate motion, the number of passes, the mass
of a single portion of powder, and the exposure time of a given surface section. It is noted that experimental
conditions can exert a significant effect on the consequences of the high-speed interaction of particles with a
substrate. Relations are suggested that allow one to correct the results of deposition efficiency determined
experimentally and to avoid mistakes in interpreting the data obtained.

Keywords cold spray, deposition efficiency, kinetics
1. Introduction
One of the most important characteristics of cold gas-
dynamic spray (CGDS), as well as of any other method of pow-
der spray, is deposition efficiency. Investigation of its depen-
dence on the governing parameters of the process (e.g., jet
temperature, velocity, and size and concentration of particles)
allows one to understand the nature of the cold spray phenom-
enon and, as a result, to reveal how the basic coating properties
are influenced by spray parameters. It also helps in controlling
these properties in accordance with necessary requirements. The
deposition efficiency is usually determined as the ratio of the
mass gain of the substrate during its exposure to the flow with a
proper set of parameters and the decrease in powder mass in the
feeder during the same time.
For many reasons, it is practically impossible to obtain a de-
position efficiency that is equal to unity. First, polydisperse
powders are usually used. As the jet during its impingement is
spreading along the substrate surface, the finest particles either
do not reach the surface at all or impact at acute angles, which
deteriorates particle attachment. Although the largest particles
are incident at a close-to-normal angle (i.e., the angle between
the particle impact velocity and the substrate surface is close to
90°), their velocities may be insufficient for particle attachment.
Such a behavior of particles can be explained by the gas-
dynamic nature of their motion. The fact is that the gas-dynamic
tract of the facility can be adjusted for spray only for a certain
interval of particle sizes (e.g., Ref 1). Particles whose size sig-
nificantly differs from the optimal value inevitably have less fa-
vorable conditions for attachment. Another reason for the de-
crease in powder deposition efficiency is that the particle
velocity at the jet periphery can be lower than it must be for
particle attachment. In addition, if the velocity is not sufficiently
high, the surface should be self-activated by the impacted par-
ticles (e.g., Ref 2, 3). This means that the first particles impact-
S.V. Klinkov and V.F. Kosarev, Institute of Theoretical and Applied
Mechanics, Siberian Branch of Russian Academy of Sciences (ITAM
SB RAS), Novosibirsk, Russia. Contact e-mail: vkos@itam.nsc.ru.
ing the natural surface always bounce. This can be avoided,
however, by ensuring high-quality preparation of the surface
and powder recirculation (reusing); a certain increase in deposi-
tion efficiency can be reached.
Due to the sufficiently complicated nature of CGDS, it is
rather difficult to measure the deposition efficiency. Generally,
three main stages of the spray process may be identified. At the
initial stage, some time is required for surface preparation (i.e.,
the induction time), when only erosion occurs without any de-
position. At the second stage, a thin layer of the particle material
(hereinafter referred to as the first layer) is formed on the sub-
strate surface. This stage is characterized by the interaction of
particles with the substrate surface, and it depends on the prepa-
ration level and properties of the surface material. The third
stage, which can be conventionally called the build-up stage, is
characterized by the growing thickness of the coating layer. In
this case, the particles interact with the surface formed by pre-
viously incident particles. Thus, it is clear why there is some
uncertainty in measuring the powder deposition efficiency.
This article has the following structure. First, sections 2 and 3
describe, respectively, the experimental setup used and the ex-
perimental evidence of the existence of an induction time in cold
spray. The idea that the transition to deposition occurs within a
certain range of particle velocities is put forward. After that, an
analytical model of coating-formation kinetics is constructed in
sections 4, 5, and 6. With the help of this model, the behavior of
the deposition efficiency is demonstrated in section 7. Proce-
dures that are commonly used to measure the deposition effi-
ciency are described in section 8, and the accuracy of determin-
ing the deposition efficiency is assessed. Experimental data on
deposition efficiency obtained by one of the commonly used
method are presented in section 9, and their correction is made
using experimental data on the induction time. It is also shown
that the use of the experimental method can lead to significant
errors. Finally, a technique for obtaining more correct data on
deposition efficiency and cold spray kinetics are proposed in
section 10.
2. Experimental Setup
The data obtained from experiments described in Ref 2 are
used in the present article. The experimental setup consists of

364—Volume 15(3) September 2006 Journal of Thermal Spray Technology

typical cold spray equipment: a thin supersonic nozzle with a proper velocity, temperature, and concentration. At the time t =

rectangular cross section; and the flow of air and helium mix-
tures allow accelerating particles up to an impact velocity suffi-
cient for deposition. The stagnation temperature and pressure
are 300 K and 2.0 MPa, respectively. The particle velocity was
calculated by the method described in Ref 1 that was previously
validated by experimental testing with the aid of a laser Doppler
anemometry and track method (Ref 2). Only data obtained from
experiments with aluminum particles (mean diameter 30 µm)
deposited onto a copper substrate were used in the present ar-
ticle.
3. Induction Time (Delay) of Deposition
To explain what the induction time is, it is necessary to see
the experimental procedure. Let the gas jet have particles with a
Peer Reviewed
0, the examined section of the substrate surface is placed into the
jet. First, the particles are not attached, they are bounced, thus
cleaning and deforming the surface. At the time t = ti, the par-
ticles begin to attach in an avalanche-like manner, rapidly form-
ing a coating. This time ti is called the induction time (or the time
of deposition delay). It characterizes the process of preparing the
surface for deposition (i.e., the conditions under which the next
particles will be attached). Experimental evidence is shown in
Fig. 1: a few particles are attached to the surface after activation
by the impact of particles. The photograph in Fig. 1 shows the
instant just before the coating starts to grow. The induction time
is related to the particle concentration in the flow. To find this
relation, it is assumed that the particle size does not significantly
differ from the average size (i.e., the case of a monodisperse
powder is considered). Note that, in the case of polydisperse
powder motion, which is the most typical one in practice,

Nomenclature

Latin Notations ṅp number of particles falling onto a unit of a surface per a
unit of time
D diameter of the cone nozzle outlet section p1 probability of a particle attachment on a free substrate
Dp diameter of the contact between a particle and the surface
substrate P2 probability of particles attaching to the surface formed by
dp diameter of a particle the particles themselves
H cross size of an outlet nozzle section directed Sex area of the surface exposed (i.e. a surface whereon the
perpendicular to a nozzle motion with respect to the particles have fallen)
substrate Sc contact area between a particle and the substrate
H size of the outlet nozzle section directed parallel to the sc contact area divided by the area exposed sc = Sc/Sex
nozzle motion with respect to the substrate Sfr free (i.e. without particles attached) surface area
hc coating thickness sfr free area divided by the exposed area sfr = Sfr/Sex
hp a typical particle thickness after deformation t time
kd deposition efficiency kd = mc/mp t0 typical time of a coating first layer formation
kd0 stationary value of the deposition efficiency kd0 = tI induction time (delay) of spraying
dmc/dmp tex time of the surface exposition
kp coefficient of a powder packing V̇p volume of the powder spent from the feeder per a unit of
L spraying strip length equal to the product of the time
substrate motion velocity with respect to the nozzle, vcr first critical velocity
and run-time of the powder feeder vcr*second critical velocity
mc mass of coating (i.e., mass of all attached particles) vp velocity of a particle impact
mc0 mass of a coating first layer vw speed of a substrate motion with respect to the nozzle
mp mass of a powder spent from the feeder. x coordinate along a nozzle motion with respect to the
mp0 mass of a powder spent during formation of the coating surface
first layer y coordinate across a nozzle motion with respect to the
mpi mass of a powder spent during the induction time surface
ṁp powder-mass flow rate from the feeder ym half-width of a strip sprayed
Nc number of the attached particles
Ṅp a particle flow rate of the powder spent from the feeder Greek Notations
(number of particles spent from the powder feeder per
a unit of time) ␣ average number of impacts into a definite point of the
Np number of particles falling onto the whole exposed surface per a unit of time
surface during exposition ␧p final degree of a particle deformation (a final particle
Npre average number of preliminary impacts into a definite strain)
surface point ␤I dimensionless parameter ␤i = (mpivw)/(ṁpR)
nc number of particles attaching to a unit of a surface ␳p density of a particle material
npi number of particles falling per a unit of a surface ␻ exponent with value 1 for the nozzles with rectangular
during the induction time sections and 1.35 for the round-section nozzles

Journal of Thermal Spray Technology Volume 15(3) September 2006—365


Peer Reviewed
Fig. 1 Copper substrate surface after a 25 s delay in a two-phase jet.
The impact velocity of aluminum particles is 600 m/s.
sufficiently narrow fractions characterized by their average size
can be conventionally distinguished. Considering each fraction
separately, information can be obtained about the powder as a
whole. In the experiment, usually a volumetric or mass flow rate
of the powder is measured. The relationship between the volu-
metric V̇p and the countable Ṅp flow rate of the powder is deter-
mined by the formula:
1 ␲d 3p
V̇p = Ṅ (Eq 1)
kp 6 p
where a typical value of the powder packing coefficient kp is of
the order of 0.5. The relationship between the mass and count-
able flow rate is determined as:
␲d 3p
ṁp = ␳p Ṅ (Eq 2)
6 p
Knowing the total particle flow loss from the powder feeder,
the particle flow per unit surface area can be easily found. For
this, let the particles be uniformly distributed along the substrate
area of the exposed surface (i.e., the average number of impacts
onto each point of the exposed surface is the same). Note that
this can be achieved rather seldom in the experiment, because
the particle concentration at the jet edges is usually lower; if the
shock-wave picture generated under off-design (nonisobaric)
conditions of jet exhaustion is highly developed, it is very diffi-
cult to predict the level of nonuniformity of the particle distri-
bution over the surface. In the general case, one should use some
additional methods or assumptions for determining the particle
flow density in the vicinity of a particular surface point. Thus,
assuming that the surface area exposed Sex coincides with the jet
cross-sectional area, and, moreover, with the nozzle-exit sec-
tion, the particle flow per unit of surface area can be easily esti-
mated as ṅp = Ṅp/Sex. Obviously, a particular point on the surface
is subjected to impact only by those particles whose centers are
within the area with a diameter Dp around this point. Here Dp is
366—Volume 15(3) September 2006
the diameter of the contact zone between the particle and the
surface, which is determined by the particle diameter and its de-
formation while impacting the surface obtained, for example, by
the following formula (from Ref 4, 5):
2dp
Dp = (Eq 3)
公3 ⭈ 共1 − ␧p兲
Assuming that the number of impacts is subjected to Pois-
son’s distribution, one can determine the probability that exactly
m particles will impact into a given point during the time t by
using the formulas taken from Ref 6:
共␣t兲m −␣t ␲D2p Sc
pm = e ␣ = Ṅp = Ṅp = Ṅp sc (Eq 4)
m! 4Sex Sex
Here ␣ is the average number of impacts into a given point
of the surface per unit time (i.e., the average impact frequency),
Sc = (␲D2p)/4 is the area of the “particle-substrate” contact, and
sc = (Sc/Sex) is the area of the “particle-substrate” contact divided
by the area exposed.
Poisson’s distribution can be approximated (if the number of
expected impacts into the point during exposure ␣tex is suffi-
ciently large) by the normal distribution. Thereby, the dispersion
(i.e., standard deviation squared) shall be equal to an average
value. Knowing the average frequency of impacts onto the point,
the average number of impacts during the induction time can be
determined by the formula:
Npre = ␣ti (Eq 5)
Because the number of impacts onto a given point of the sur-
face must be approximately equal, and it characterizes the acti-
vation processes, an assumption that the delay time of deposition
is inversely proportional to the powder flow rate follows from
Eq 4 and 5.
The delay of time makes conducting the experiment more
difficult. Thus, pulse (short work time) regimes and regimes
with a moving substrate are most often used in the experiment.
Obviously, if the particle concentration in the flow is insufficient
or the exposition time is short, it can be mistakenly concluded
that the powder tested is suitable for CGDS. Thus, the induction
time of the process must be measured to choose a correct oper-
ating time for the device and a correct particle concentration for
the powder in the flow. In the second case, a correct velocity of
substrate motion should be chosen. In this case, the exposure
time of the unit (or point) of the surface is calculated by the
formula:
h
tex = (Eq 6)
vw
where h is the size of the nozzle outlet directed toward the sub-
strate motion with a velocity vw. Note that this value is a variable
for round-cut nozzles and coincides with the diameter only at the
axis.
Figure 2 shows the induction time as a function of the impact
velocity obtained from experiments on spray aluminum particles
Journal of Thermal Spray Technology
 ability of particle attachment to the occupied surface). This im-

Peer Reviewed
plies that the total number of particles attached is determined by
the expression:

dNc =
p1sfr + p2共1 − sfr兲 dNp (Eq 8)

To evaluate the first layer mass, some additional simplifica-

tions are made. Consider that the probability of attachment to the
free surface p1 at t ⱕ ti is equal to zero. As the coating thickness
increases, let the probability of attachment to the free surface p1
not vary during the time that the first layer is formed. This allows
the authors to obtain analytical solutions (Eq 9 and 10) for the
free surface and mass of the particles deposited mc, respectively:

Fig. 2 Induction time versus the impact velocity for aluminum par-
sfr = 再 1
exp关−p1␣共t − ti兲兴
t ⬍ ti
t ⱖ ti
(Eq 9)

ticles. The mass flow rate of the powder per unit area is 0.06 kg/m2/s.
t ⬍ ti

冦
0
mc = p1 − p2 ṁp
on a copper substrate. An approximating curve of the ti = p2ṁp共t − ti兲 + 关1 − exp共−p1␣共t − ti兲兲兴 t ⱖ ti
a[(1/√vp −vcr) − (1/√v*cr − vcr)] type with the values a ≈ 365 s, vcr p1 ␣
= 550 m/s, and v*cr = 850 m/s is used. (Eq 10)
The picture demonstrates a significant growth for the induc-
tion time with decreasing impact velocity. In practice, particle The first layer formation is practically completed when con-
attachment is observed in this case during some arbitrarily large dition 11 is fulfilled. Further, as it follows from Eq 10, the coat-
period of exposure time. A certain conventional critical velocity ing increases linearly (i.e., dmc/dt = Const):
may be identified, such that the surface is eroded if the particle
impact velocity is lower than the critical value. If the velocity is
1
higher than the critical value, the particles can attach to the sur- t0 − ti ≈ 3 (Eq 11)
face if a sufficient level of the surface activation is reached. The p1␣
following fact can be also noted: with increasing impact veloc-
ity, the delay time decreases to zero. From the physical point of Substituting t0 from Eq 11 into Eq 10 instead of t, Eq 12 is
view, this means that the particles can attach to a nonactivated obtained for the particle mass in the first layer of the coating mc0.
surface. Thus, the second critical velocity can be identified, such Here ␣ is also replaced using Eq 3 and 4, where ␧p = 1−hp/dp:
that the particles attach to the natural (i.e., nonactivated) surface

冉 冊 冉 冊
if the impact velocity is higher than the second critical velocity.
ṁp p2 ␳pdp p2
mc0 = 1+2 = Sex 共 1 − ␧p 兲 1 + 2 (Eq 12)
␣ p1 2 p1
4. Determination of the Coating First
Layer Mass Note that the number of particles in the first layer (and, con-
sequently, the first layer mass) depends on probabilities, and,
The following assumptions can be applied for determining consequently, on the properties of the substrate and particle ma-
the first layer mass (i.e., the mass of the layer formed on the terials.
surface when the substrate is completely coated at least by one When the probabilities are commensurate in value (if the ma-
layer of particles). Assume that, at a certain time, some part of terials of the substrate and particles are identical, the probabili-
the exposed surface with the area Sex, area Sfr, is free (i.e., it is not ties must be identically equal), the particle mass in the first layer
occupied by particles). Then, introduce the free area divided by can be evaluated as:
the exposed area sfr = Sfr/Sex. During the next moment of time,
only a few particles, dNp, occur on the exposed surface. Among
mc0 ␳pdd
them, sfrdNp particles are on the free area. Among that number of ≈3 共1 − ␧p兲 (Eq 13)
particles, p1sfrdNp particles are fixed on the free surface (p1 is the Sex 2
probability of attachment on the free surface). Thereby, the free
area occupation is determined by the expression: The value of the final strain of a particle can be taken as approxi-
mately 0.5 (Ref 7).
dsfr = −p1sc sfr dNp (Eq 7) Owing to the more sophisticated nature of cold spray, the
probabilities introduced are effective average values; the experi-
On the other hand, during the same time, p2(1−sfr)dNp par- mental kinetics curve corresponding to these probabilities can be
ticles will be attached onto the occupied surface (p2 is the prob- approximated by Eq 10.

Journal of Thermal Spray Technology Volume 15(3) September 2006—367

 5. Buildup Stage of Spray
Peer Reviewed

Obviously, after the formation of the first layer of the coating,

the curve of the further increase in the coating mass should have
a constant slope, because the particles interact with the surface
that they themselves have formed. A decrease or increase in the
growth rate should be considered as evidence of changes in in-
teraction conditions with the surface. Assume that the first layer
of mass mc0 is formed on the substrate surface at the time t = t0.
Then, a further increase in the coating mass (or the number of
particles attached) is described by the linear dependence:

mc = mc0 + kd0ṁp共t − t0兲 = mc0 + kd0共mp − mp0兲, t ⬎ t0 (Eq 14)

On the other hand, assuming that the exponent in Eq 10 is

negligibly small compared with unity, the following expression
is derived:

p1 − p2 ṁp
mc = p2ṁp共t − ti兲 + at t ⬎ t0 (Eq 15)
p1 ␣

Under the following assumptions, Eq 15 acquires the form of

Eq 14:

p2 = kd 0 (Eq 16)

3mc0
p1 = − 2kd0 (Eq 17)
mp0 − mpi

6. Kinetics of Growth of the Coating

Mass
Fig. 3 Coating mass versus the mass of the powder spent for (a) iden-
Figure 3 demonstrates the variation in the mass of the depos- tical and (b) different induction times. See the description in the text.
ited layer versus the powder mass spent (or the time at constant
powder consumption from the feeder). The mass of the powder
spent is plotted on the x axis, and the coating mass is plotted on duction time is larger. This variant is shown in Fig. 3(b). Obvi-
the y axis. In Fig. 3(a) and (b), the probability of particle attach- ously, in this case the coating growth is additionally delayed.
ment to the occupied surface (i.e., at the build-up stage of the
spray) is assumed to be equal to 0.5. Particle attachment to the
free surface is somewhat better (case 1), somewhat worse (case 7. Deposition Efficiency
2), and equal (case 3) compared with particle attachment to the
occupied surface. A straight line from the coordinate origin cor- The next step is to determine the experimental deposition ef-
responds to the case where the induction time is equal to zero ti ficiency kd as a ratio of the coating mass to the mass of the pow-
= 0 (or mpi = 0), and the probabilities are p1 = p2 = 0.5 (i.e., the der spent. At the build-up stage of spray, the derivative dmc/dmp
surface and particle materials are identical). Curves 1 and 2 refer is assumed to be kd0, and it is called the theoretical deposition
to case 1 (p1 > p2) and case 2 (p1 < p2). The induction times efficiency.
shown in Fig. 3(a) are identical. Figure 3(b) presents the case Figure 4 shows the dependence of the modeled deposition
where the induction times are different ti1 < ti2 (or mpi1 < mpi2). efficiency on the mass of the powder consumed. Curves 1 and 2
The solid curves are built by Eq 10 for cases 1 and 2, respec- that were calculated by Eq 18, derived from Eq 10, correspond to
tively. As mp increases from zero to mpi, the coating mass is cases 1 and 2 considered above at different induction times:
equal to zero, it further increases according to Eq 10, and, at mp
> mp0, it approaches the approximating curve determined by Eq
15. The dashed curve 3 refers to case 3 (p1 = p2 = 0.5). Obvi- 冉
kd = kd0 1 −
mpi
mp 冊
再 冋 册冎
ously, the coating grows more rapidly in case 1 where the par-
ticle attachment to the free surface is somewhat better than at- mc0 − kd0共mp0 − mpi兲 3共mp − mpi兲
+ 1 − exp −
tachment to the occupied surface. When attachment to the free mp mp0 − mpi
surface is somewhat worse, it is logically expected that the in- (Eq 18)

368—Volume 15(3) September 2006 Journal of Thermal Spray Technology


Fig. 4 Deposition efficiency versus the mass of the powder spent. See
the description in the text.
Expression 18 is simplified to Eq 19 if the second component
in Eq 18 is not taken into account (i.e., the materials of particles
and the substrate are identical). Curves 3 and 4 are constructed
by Eq 19, and correspond to cases 1 and 2.
mp ⬍ mpi
冦冉 冊
0
mc
kd = = mpi (Eq 19)
mp kd 0 1 − mp ⱖ mpi
mp
Note that the influence of the substrate material on deposition
kinetics is described with the help of the induction time and at-
tachment probability on the free surface. Comparing curves 1
and 3, as well as curves 2 and 4, in Fig. 3 and 4, one can conclude
that deviation of the probability of attachment on the free surface
from that on the occupied surface affects the deposition kinetics
at the initial part of coating growth. In the first approximation,
only the induction time can be taken into account to explain the
deposition kinetics, and the difference between the attachment
probabilities on the free and occupied surfaces can be neglected.
In Fig. 3, it corresponds to the replacement of the kinetics curve
by a curve whose slope equals the theoretical deposition effi-
ciency. However, as seen in Fig. 3, the “effective” induction
time appears to be unequal to the “true” induction time. Because
the experimental difference between the “effective” and “true”
induction times can be expected to be insufficient, a further
analysis will be performed using Eq 19.
For any number of particles spent from the feeder, there is
always a certain error in determining the theoretical deposition
efficiency. The larger the mass of the powder spent (spray time),
the closer the deposition efficiency experimentally measured to
the theoretical one:
| kd − kd 0 | mpi
= (Eq 20)
kd 0 mp
It is seen from Eq 19 that the experimental value tends to zero
because the number of particles spent tends to the number of
Journal of Thermal Spray Technology
particles that is necessary for surface activation, but not because
Peer Reviewed
the theoretical deposition efficiency is equal to zero. Thus, to
explain the phenomenon correctly, it is important to know the
induction time of deposition.
8. Measurement Procedures
Three methods of the experimental setup are commonly used
to measure the deposition efficiency. Each method will be con-
sidered in detail.
8.1 Method A
The substrate is inserted into the two-phase jet fast enough,
and then it stands still during a certain exposure time. After a
proper exposure time, the mass of the powder spent from the
feeder and the mass growth of the substrate are measured. In this
case, the powder flow rate from the feeder is constant in time, as
a rule (i.e., ṁp = Const). Thus, a relationship between the experi-
mental and theoretical deposition efficiencies is determined ac-
cording to the formula:
冉
kd = kd0 1 −
ti
texp 冊 (Eq 21)
Note that the applicability of this method is restricted by
time, because the coating increases nonuniformly. As a rule, it
has a conical shape, which leads to variation in the conditions of
particle attachment at the jet edges. This degrades the measure-
ment accuracy of the experimental and theoretical deposition
efficiencies. It can be expected that this method is less accurate
compared with two others considered below.
8.2 Method B
A certain dose of the powder is placed into a cartridge. At the
zero time (t = 0), the cartridge is pushed into the gas-dynamic
path, so that a portion of the powder goes out as a dense cloud.
Thus, a pulsed spray regimen with a rather high particle concen-
tration near the substrate surface takes place. The substrate mass
is determined before and after the exposure. In this case, Eq 19
can be applied. As with method A, and for the same reason, this
method is restricted by the powder dose placed into the car-
tridge. However, it should be expected that a preliminary weigh-
ing of the powder portion would provide better accuracy than
that in method A.
8.3 Method C
The substrate moves under the jet with a constant velocity.
Therewith, it can cross the same place on the surface more than
one time. In practice, the substrate velocity and the number of
passes are adjusted to produce a sufficient coating mass for the
measurement accuracy. Here, the flow rate of the powder spent
from the feeder is also constant in time.
To derive an expression for deposition efficiency, the el-
ementary area is considered as dxdy. Obviously, moving under
the jet toward x, it is subjected to the impacts of particles during
the time h(y)/vw. Thus, the number of particles incident on the
elementary area is ṅph(y)/vw)dxdy. If the number of incident par-
ticles is greater than the number necessary for surface activation,
which is npidxdy, then the number of particles attached to
Volume 15(3) September 2006—369
 the elementary area dxdy is ncdxdy = p2(ṅph(y)/vw − npi)dxdy.
Peer Reviewed

Thus, the total number of particles spent is determined by the

integral:

ṅph共y兲
兰 dx 兰
L +H Ⲑ 2
Np = dy (Eq 22)
0 −H Ⲑ 2 vw

Expression 22 shows that the number of particles spent dur-

ing the feeder operation time L/vw is Np = ṄpL/vw, where L is the
length of the spray strip, taking into account the number of
passes. By virtue of the simplicity of the physical interpretation,
integration is not necessary.
The total number of particles attached is determined by the
expression:

兰 dx 兰
L +ym
Nc = p2共ṅph共y兲 Ⲑ vw − ni兲dy (Eq 23)
0 −ym

Here, the half-width of the deposited strip ym is determined

from the condition npi = ṅph(ym)/vw or ti = h(ym)/vw.
The deposition efficiency measured in the experiment is de-
termined by the ratio of Eq 23 and 22. In the case of a plane
nozzle, the relations are easily integrated because the exposure
time of the elementary area of the surface is independent of the
coordinate y. The following expression is obtained for the depo-
sition efficiency in this case:
Fig. 5 Deposition efficiency measured in the experiment and the cor-
rected value: experimental values (1), corrected values (2), and the val-
ues that would be obtained in an experiment with a weight dose of 17 mg
(3).
In the case of a round nozzle, the diameter of the outlet section
is used instead of h. Because the value in brackets is smaller than
unity, the experimental deposition efficiency in the case of the

冉
k d = kd 0 1 −
mpivw
ṁph 冊 冉
= kd 0 1 −
tivw
h 冊 (Eq 24)
nozzle with a rectangular outlet is closer to the theoretical value.

9. Correction of the Deposition

Note that, considering Eq 6, Eq 24 can be derived from Eq 19
and 21.
In the case of a conical nozzle, the expression is complicated
because the size h is changed according to the formula:
h共y兲 = 2公共D Ⲑ 2兲2 − y2
(Eq 25)
where D is the jet diameter and y is the coordinate perpendicular
to the direction of substrate motion counted from the centerline
of the coating strip.
In this case, the deposition efficiency is determined by the
formula:
2 tivw
k d = kd 0 共arcsin公1 − ␤2i − ␤i公1 − ␤2i 兲, ␤i =
␲ D
(Eq 26)
The function on the right side of Eq 26 can be replaced by a
more simple approximate expression (1−␤i)1.35. Therefore, an
extended (more common) formula can be accepted for the depo-
sition efficiency, where ␻ is equal to 1 or 1.35 for the rectangular
or round nozzle cross sections, respectively:
Efficiency
Using the data presented in Fig. 2, one can estimate the pow-
der mass that will be spent for surface activation and first layer
formation. In the experiments on determining deposition effi-
ciency, method B was used, where the powder weight dose was
equal to 100 mg, which yielded a specific mass of about 3.3
kg/m2. Substituting the values found into Eq 19, a correction can
be found of the theoretical deposition efficiency for the data ob-
tained in the experiment (Ref 2) (Fig. 5). It is seen that the dose
with a weight of 100 mg is sufficient for obtaining reliable ex-
perimental data. Nevertheless, if the weight of the dose is much
smaller than 100 mg (e.g., Fig. 5 shows the data that would be
obtained with a dose of 17 mg), the differences would be fairly
noticeable; for instance, the critical velocity would be 700 m/s
rather than the real value of 600 m/s.
10. Discussion
Using the theory of deposition kinetics that was described
above, another procedure can be proposed for the experimental
determination of the deposition efficiency. Namely, it is neces-
sary to measure the coating mass during deposition twice. First,
the coating mass is measured as usual, but then the coating mass

冉 冊
is measured with the increasing mass of the powder spent from
mpivw ␻
kd = kd0 1 − (Eq 27) the feeder. Then, the theoretical deposition efficiency is deter-
ṁph mined directly using the formula:

370—Volume 15(3) September 2006 Journal of Thermal Spray Technology

 mc2 − mc1
kd 0 ≅ (Eq 28)
mp2 − mp1
Moreover, it is possible to obtain an “effective” induction
time using this procedure. For this purpose, it is necessary to find
a point of intersection between the x axis and the line constructed
through two experimental points obtained: (m p1 ,m c1 ) and
(mp2,mc2). This intersection point yields the “effective” induc-
tion mass (induction time). Obviously, for a detailed insight into
the deposition kinetics, the deposition efficiency should be mea-
sured at many points.
11. Conclusions
The article considers the methods of measuring the deposi-
tion efficiency and evaluating the measurement accuracy in cold
spray. The authors place the primary emphasis for estimating the
effect of the delay time on the accuracy of the measurement of
deposition efficiency. It is shown that the values experimentally
observed are affected by experimental conditions, such as the
velocity of substrate motion, the number of passes, the mass of
the single powder portion, and the exposure time of a given sur-
face section. If the parameters are chosen incorrectly, erosion is
observed instead of deposition, and one may wrongly conclude
Journal of Thermal Spray Technology
that there is no cold spray produced under the parameters con-
Peer Reviewed
sidered (i.e., at the impact velocity and the particle size for the
materials chosen for the substrate and particles). The present ar-
ticle suggests relations that allow one to correct the results of the
deposition efficiency determined experimentally and to avoid
mistakes in interpreting the data obtained.
References
1. A.P. Alkhimov, S.V. Klinkov, and V.F. Kosarev, The Features of Cold
Spray Nozzle Design, J. Therm. Spray Technol., 2001, 10(2), p 375-381
2. A.P. Alkhimov, V.F. Kosarev, and A.N. Papyrin, A Method of Cold Gas
Dynamic Deposition, Sov. Phys. Dokl., 1990, 35(12), p 1047-1049
3. S.V. Klinkov and V.F. Kosarev, Influence of Surface Activation on
Cold Spray Process, Physical Mesomechanics, 2003, 6(3), p 85-90, in
Russian
4. S.V. Klinkov and V.F. Kosarev, Modeling of Particle-Substrate Adhe-
sive Interaction under Cold Spray, Physical Mesomechanics, 2002,
5(3), p 27-35, in Russian
5. S.V. Klinkov, V.F. Kosarev, and A.N. Papyrin, Modeling of Particle-
Substrate Adhesive Interaction Under the Cold Spray Process, Thermal
Spray 2003, Advancing the Science and Applying the Technology, B.
Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM Inter-
national, 2003, 1, p 27-35
6. G.A. Korn and T.M. Korn, Mathematical Handbook for Scientists and
Engineers, McGraw Hill Book Company, New York, 1968
7. A.P. Alkhimov, S.V. Klinkov, and V.F. Kosarev, Experimental Study of
Deformation and Attachment of Microparticles to an Obstacle Upon
High-Rate Impact, J. Appl. Mech. Tech. Phys., 2000, 41(2), p 245-250
Volume 15(3) September 2006—371